Conference paper

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLI-B4, 2016
XXIII ISPRS Congress, 12–19 July 2016, Prague, Czech Republic
QUALITY TEST VARIOUS EXISTING DEM IN INDONESIA
TOWARD 10 METER NATIONAL DEM
Fahmi Amhara
a
Geomatics Research Center, Geospatial Information Agency
Jl. Jakarta-Bogor Km. 46 Cibinong INDONESIA – fahmi.amhar@big.go.id
Commission IV
KEY WORDS: Quality Test, DEM, Indonesia,
ABSTRACT:
Indonesia has various DEM from many sources and various acquisition date spreaded in the past two decades. There are DEM from
spaceborne system (Radarsat, TerraSAR-X, ALOS, ASTER-GDEM, SRTM), airborne system (IFSAR, Lidar, aerial photos) and also
terrestrial one. The research objective is the quality test and how to extract best DEM in particular area. The method is using
differential GPS levelling using geodetic GPS equipment on places which is ensured not changed during past 20 years. The result
has shown that DEM from TerraSAR-X and SRTM30 have the best quality (rmse 3.1 m and 3.5 m respectively). Based on this
research, it was inferred that these parameters are still positively correlated with the basic concept, namely that the lower and the
higher the spatial resolution of a DEM data, the more imprecise the resulting vertical height.
1. INTRODUCTION
from processing DSM so that its height just above the ground.
Contours contained in the topographic maps generated from the
data extraction using DTM (Petrie & Kennie, 1987).
Mapping of Indonesian Topography (RBI) is one of the main
tasks of the Geospatial Information Agency (BIG). BIG is
assigned to map RBI from small scale to large scale. One of the
main elements in the map is hypsography which will describe
the shape of the earth's surface. The earlier primary data
sources are aerial photographs.
DTM formation requires advanced process to lower the height
of the surface to the terrain. Indonesia conventional
topographical mapping using stereo plotting techniques by
forming a 3D model of stereopair of aerial photographs or
orthorectified radar image (ORRI) and its stereomate formed
from the ORRI and DSM. Through this 3D model we creates
hypsography elements like mass point, ridge line (break line),
and the river's which shape DTM and contour (Li, Zhu, & Gold,
2005).
1.1 Background
Along with the development of remote sensing technology, it is
now available a wide range of data that can be used to obtain
elevation data digitally, known as Digital Elevation Model
(DEM). Some of this data has also been used by BIG in
mapping the RBI. Some types of data are SRTM DEM, IFSAR,
TerraSAR-X, ASTER-GDEM and LIDAR. The DEM data have
been used for mapping on a scale of 1: 50,000 and 1: 25,000.
Thus, the challenge today is on large-scale mapping, which is a
scale of 1: 10,000, 1: 5,000, and even 1: 1000, which has been
awaited by local government agencies and the public for
detailed spatial plan. Large scale mapping would require DEM
data with higher resolution and accuracy level anyway. This
hi-res DEM is also required to ortho-process high-resolution
satellite imagery (CSRT).
The research question is how the real-accuracy of each DEM is
measured and how to integrate the entire DEM becomes a
national DEM. A consolidated national DEM is an essential
key to implement One Map Policy.
1.2 DEM
DEM data may be Digital Surface Model (DSM) and Digital
Terrain Model (DTM). DSM covers the entire surface of the
object (buildings, vegetation, etc.), while DTM represents the
height above ground level (bare earth). DSM presents a surface
that is obtained when the acquisition while DTM generated
Currently, there are some well known global DEM available:
SRTM, ASTER-GDEM, and TerraSAR-X.
SRTM mounted on the Space Shuttle and the earth's surface
elevation data obtained with synthetic aperture radar (SAR)
technology. SRTM utilizes radar interferometry to process data
until a DEM. In radar interferometry, two radar images were
acquired simultaneously required by two antennas. Two images
that have these differences allow for the calculation of the
elevation of the earth's surface. SRTM flown for 11 days at 11
to 22 February 2000, and collected data more than 80% of the
earths land 60°N to 56°S. On SRTM there are two types of
antenna panels, C-Band and X-Band. The C-Band will be
processed by Jet Propulsion Laboratory (JPL), which provides
global DEM and distributed through the USGS. X-Band will be
processed and distributed by the German Aerospace Center
(DLR), which resulted higher resolution DEM, but it does not
have global coverage. Data C-Band is expected to have an
accuracy of horizontal and vertical approach the 20 m and 16 m
(error linear at 90% confidence), to release final data 1 arc-sec
of the United States (NASA, 2005).
Vertical precision of SRTM DEM depends on the phase noise
in radar, while the horizontal resolution depends on the ratio of
This contribution has been peer-reviewed.
doi:10.5194/isprsarchives-XLI-B4-111-2016
111
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLI-B4, 2016
XXIII ISPRS Congress, 12–19 July 2016, Prague, Czech Republic
signal-to-noise as a function of horizontal wavelength.
Horizontal resolution of SRTM data is divided into: 1 arc-sec or
30 m (SRTM30) and 3 arc-sec or 90 m (SRTM90). SRTM30
data is only available for locations in the US while the SRTM90
data is available to the public free of charge (Gesch, 2005).
stereo reconstruction. If both methods were compared, SAR
interferometry is more accurate than radargrammetry, but
radargrammetry overcome the weaknesses of interferometry
related temporal disconnection as a result of the repeat pass for
11 days (Raggam, 2010).
NASA and METI Japan then released the second version
GDEM (GDEM 2) in mid-October 2011. Some things improved
from GDEM first version, such as the horizontal resolution is
improved using a smaller kernel (5 x 5 after 9 x 9 in GDEM
1.0) as well as improvements the masking waters. GDEM 2 has
a vertical accuracy of 20 m and 30 m horizontal accuracy at
95% confidence level. ASTER GDEM distributed in GeoTIFF
format with geographic coordinate system. Horizontal datum
used is WGS84, while the vertical datum is EGM96 (Athmania,
2014; Li, 2013; Forkuor, 2012)
1.3 Accuracy of DEM Data
Accuracy DEM determined from the error in the spatial
coordinates (X, Y, Z) points tested on the DEM. Validation
accuracy DEM data can be done by comparing DEM tested and
DEM referenced or by testing the amount of vertical errors of
DEM using Ground Control Points (GCP) are obtained from
field surveys (Mukherjee, 2013).
Indonesia has published a regulation about geometric accuracy
of national topographic map (RBI) as shown in Table-1 and 2.
TerraSAR-X is the first remote sensing satellite implemented in
a public-private partnership business model and is the follow-up
mission of SRTM. TerraSAR-X can be operated in spotlight,
stripmap, and scanSAR-mode with two polarizations in some
combination. Each mode has a different resolution. Stripmap
mode has a resolution of 3 m x 3 m, spotlight mode 1 m x 1 m,
while the scanSAR mode has a resolution of 16 m x 16 m ((Li,
2013).
Base map position accuracy value in Table 2 is the value for the
accuracy Circular Error 90% (CE90) and Least Error (LE90)
horizontal to vertical accuracy, which means that the position
error does not exceed the value of the base map accuracy with
90% confidence level.
CE90 and LE90 value obtained by the following formula:
CE90 = 1.5175 x RMSEr
LE90 = 1.6499 x RMSEz
** RMSE = Root Mean Square Error at position
DEM formation of TerraSAR-X Data can use interferometry or
radargrammetry. Interferometry uses the phase difference of
the two images with the same geometry, the repeat pass for 11
days or single pass simultaneously. The radargrammetry use
two images with different geometry or the acquisition of a 3D
Table 1. Geometric Accuracy of RBI maps.
map accuracy (m)
no
Scale
contourinterval
(m)
Class 1
Horizontal
(CE90)
Class 2
Vertical
(LE90)
Horizontal
(CE90)
Class 3
Vertical
(LE90)
Horizontal
(CE90)
500
Vertical
(LE90)
1.
1:1,000,000
400
200
200
300
300.00
500.00
2.
1:500,000
200
100
100
150
150.00
250
250.00
3.
1:250,000
100
50
50
75
75.00
125
125.00
4.
1:100,000
40
20
20
30
30.00
50
50.00
5.
1:50,000
20
10
10
15
15.00
25
25.00
6.
1:25,000
10
5
5
7.5
7.50
12.5
12.50
7.
1:10,000
4
2
2
3
3.00
5
5.00
8.
1:5,000
2
1
1
1.5
1.50
2.5
2.50
9.
1:2,500
1
0.5
0.5
0.75
0.75
1.25
1.25
10.
1:1,000
0.4
0.2
0.2
0.3
0.30
0.5
0.50
Table 2. Geometrical Accuracy of RBI-maps according to Class
Accuracy
Class 1
Class 2
Class 3
Horizontal
Vertikal
0.2 mm x scale numer
0.5 x contour interval
0.3 mm x scale numer
1.5 x accuracy Class 1
0.5 mm x scale numer
2.5 x accuracy Class 1
This contribution has been peer-reviewed.
doi:10.5194/isprsarchives-XLI-B4-111-2016
112
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLI-B4, 2016
XXIII ISPRS Congress, 12–19 July 2016, Prague, Czech Republic
1.4 DEM Data Accuracy Research
1.5 BIG NAS-DEM with National Coverage
Various methods to validate the accuracy DEM data has been
done in research. The research of the DEM can be considered a
more thorough DEM reference and the other comparison to
GPS measurement data. Some of the research that validates the
accuracy of the DEM DEM by comparing with reference.
BIG as the state organization which is responsible in making
Basic Geospatial Information also required to be able to provide
elevation datasets for the whole of Indonesia (Indonesian
National Elevation Dataset / INED). However, not all types of
data available in the BIG DEM cover all of Indonesia. BIG
DEM has collected from various sources and acquisition year.
Sometimes in one location there are a variety of DEM
(Photogrammetry, SRTM30, Aster-GDEM, StereoplottingTerrasarX, IFSAR, Lidar, etc.). How to integrate the entire
DEM to be a DEM-National? DEM data available for the
whole area only SRTM90 data and Aster-GDEM. Information
DEM data availability can be seen in Table 1. With both of
these reasons, it is necessary to test the accuracy for each type
of data existing DEM (Hoja & d’Angelo, 2010).
The other research is to test the accuracy of the DEM is
compared with that of GPS measurements, such as examines the
data DSM TerraSAR-X with visual testing methods and
statistical analysis,
Others use a combination of both, as Mukherjee et al. (2013)
comparing DEM Cartosat, ASTER and SRTM, which tested the
against DGPS point data and USGS DEM, which examines
ASTER GDEM2 GPS measurements and SRTM in China
(Mukherjee, 2013).
Tabel 3. DEM Data availlability at BIG
Data
Resolution/Scale
RBI 10K (old)
1:30.000
IFSAR
Format
Captured Year
Coverage
1994
Part of West Java
.bil
1997
West Java
Foto Udara Digital
15 cm
.tiff
2014
Bogor
LiDAR
4 points/m2
las
2014
Bogor
TERRASAR-X
9m
.bil
2010-2011
Java
ASTER-GDEM
15 M
.dted
SRTM30 (30)
30m
.dted
1999-2000
Indonesia
SRTM90 (90)
90m
.dted
1999-2000
Indonesia
Indonesia
2. MATERIAL & METHODS
2.1 Locus
This research was conducted in the region around Cibinong, and
Bogor. The choice of location this study considers what
topography and accessibility as well as the range of CORS
stations BAKO as reference for GPS measurements. The
topography of the study area represent criteria that form the
plains, hills and mountains. Associated with DEM data, this
region represents an area with a low frequency, medium and
high that will affect the accuracy of the data-tia each DEM data.
In this case Cibinong region represents an area with flat
topography, Bogor represent the hills and mountains. DEM data
accuracy is determined by the topography and spatial resolution.
Distribution of GPS measurement point in this study is shown
in Figure 1.
There are 2 kinds of data, i.e. DEM data and GPS data field
measurement results. DEM data used in this study consisted of
ASTER-GDEM, IFSAR-DTM, TerraSAR-X (TSAR-X),
SRTM30, SRTM90 and Topographic maps (RBI) which is
based on Digital Aerial Photographs.
GPS measurement data with measurements performed in the
field at some point determined as described in next section.
Fig 1. Distribution of GPS test point
This contribution has been peer-reviewed.
doi:10.5194/isprsarchives-XLI-B4-111-2016
113
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLI-B4, 2016
XXIII ISPRS Congress, 12–19 July 2016, Prague, Czech Republic
2.2 Selecting GPS Test Point
The GPS Test point in the field is determined by the following
criteria:
a relative stable flat place which believed to be unchanged
in the last 20 years; in certain place a steady inclined place
is chosen.
The measurement point spread evenly, a minimum of 10
points in the selected area.
2.3 Measurement Procedure
Measurements were made using CORS GPS Levelling with
BIGs “BAKO” point as a reference. At the first measurement
several observation for 15', 30', 60 'and 120' were made to find
the optimal observation duration. It is found that duration of
15’ - 20’ gives the results that differ only 5 cm to data from
120’ duration. Standard deviation (sd-H) of the internal GPS is
still under 4 mm. So it was decided the next observation is
carried out for 30 minutes and is enough for the expected Zaccuracy. Measurements with this method is suitable for shortbaseline measurement.
This brief static method relies on the process of rapid
determination of phase ambiguity. Besides requiring the data
processing software that is reliable and fast, short static method
also requires a good satellite viewing geometry also. Accuracy
(relative) position of the point obtained in using rapid static
measurements are in the order of centimeters.
GPS data will be processed using DGPS methods (relative to
any other point) where the results tied to the fixed observation
stations CORS, which here is belonging to the Geospatial
Information Agency (BAKO site code). To process the GPS
data, a commercial software will be used.
Fig 2. Calculation on SRGI website
To obtain Z-value at the test point in each DEM, the existing
nearest cell coordinates will be extracted using ArcGIS software
tools to Extract Value Points. This Z-value is normally already
in orthometrical height. The deviation of Z-value (DEM) to the
H-orth (GPS) will be calculated. From all test points, a RMS
(σ) which indicates the accuracy of DEM will be calculated.
Gunung Picung Zvalues does not exist because the RBI in the
region are not available. The results of the processing tools in
EGM96 datum (datum adjust GDEM ASTER and SRTM) is in
Table-5.
3. RESULT & DISCUSSION
The following table-4 shows the H-orth at the test-points and its
accuracy according the geoid model. Table-5 contains the Zvalue of all test points in each DEM. And table-6 shows the
Height-differences of Z-value in each DEM to the GPS
meausred test points
2.4 DEM Accuracy Analysis
To get orthometrical height (H-orth), the measured ellipsoidal
height (H-ell) will be corrected with geoid undulation using
EGM2008-1 which is served at srgi.big.go.id and EGM1996 at
www.unavco.org. Value undulations will refer to the WGS84
ellipsoid.
Table 4. Calculation of elevation on EGM08 and EGM96
EGM08
NO
LOKASI
LATITUDE
LONGITUDE
EGM96
Δh(EGM08-
Hell
N
Hort
N
Hort
EGM96)
1
Kota Wisata
6° 22' 33.43763" S
106° 57' 30.11776" E
80.3747
18.3606
62.0141
18.566
61.8087
0.2054
2
Sentul
6° 35' 21.11698" S
106° 53' 52.37036" E
331.9196
19.124
312.7956
18.777
313.1426
-0.3470
3
Kota Kembang
6° 26' 38.99122" S
106° 49' 43.88707" E
124.4043
18.1615
106.2428
18.320
106.0843
0.1585
4
Pemda Kab Bogor
6° 28' 50.01309" S
106° 49' 27.26370" E
147.3559
18.2628
129.0931
18.348
129.0079
0.0852
5
IPB Dramaga
6° 33' 39.60649" S
106° 43' 34.20247" E
199.1913
18.1469
181.0444
18.038
181.1533
-0.1089
6
Gunung Picung
6° 40' 16.00265" S
106° 39' 49.04604" E
601.2899
18.5502
582.7397
17.801
583.4889
-0.7492
7
Citeureup
6° 29' 12.26930" S
106° 52' 43.83558" E
144.1379
18.483
125.6549
18.527
125.6109
0.0440
8
Hambalang
6° 32' 27.04034" S
106° 53' 32.79216" E
466.4407
18.8025
447.6382
18.660
447.7807
-0.1425
Katulampa
6° 37' 59.40724" S
106° 50' 12.72371" E
367.1154
18.9719
348.1435
18.661
348.4544
-0.3109
10
9
Lap. Sempur
6° 35' 29.01435" S
106° 48' 02.24921" E
249.3444
18.5842
230.7602
18.431
230.9134
-0.1532
11
BIG-2
6° 29' 28.00350" S
106° 50' 56.08653" E
157.6958
18.3878
139.3080
18.444
139.2518
0.0562
This contribution has been peer-reviewed.
doi:10.5194/isprsarchives-XLI-B4-111-2016
114
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLI-B4, 2016
XXIII ISPRS Congress, 12–19 July 2016, Prague, Czech Republic
Table 5. Z-value of all test points in each DEM
NO
ASTER
DTM
TSAR-X
SRTM30
SRTM90
RBI
1
Kota Wisata
NAMA TITIK
62.9568
54.3649
61.0940
62.7958
60.6488
59.7946
2
Sentul
317.6590
298.8840
310.1740
308.9840
308.8190
303.3267
3
Kota Kembang
105.1530
106.3980
109.9335
106.9770
113.4260
105.8116
4
Pemda Kab. Bogor
125.0700
125.2190
132.7078
133.2970
131.4480
124.9148
5
IPB Dramaga
174.8660
181.5700
182.3449
180.4360
180.2380
175.1089
6
Gunung Picung
578.7800
566.1190
586.4632
582.7360
580.4970
N/A
7
Citeureup
123.7480
124.7050
129.9050
131.5870
132.2380
127.4523
8
Hambalang
453.5790
446.9700
449.6935
453.7130
470.2700
435.6485
9
Katulampa
350.1590
343.8810
345.8359
347.8530
349.5390
345.7264
10
Lapangan Sempur
222.0770
232.5440
236.2502
234.5100
241.0770
230.1532
11
BIG-2
130.2010
137.4450
139.9438
142.4530
142.2100
134.9438
Table 6. Height-differences of Z-value in each DEM to the GPS meausred test points
NO
NAMA TITIK
ASTER
DTM
TSAR-X
SRTM30
SRTM90
RBI
1.148
-7.444
-0.715
0.987
-1.160
-2.014
1
Kota Wisata
2
Sentul
4.516
-14.259
-2.969
-4.159
-4.324
-9.816
3
Kota Kembang
-0.931
0.314
3.849
0.893
7.342
-0.273
4
Pemda Kab. Bogor
-3.938
-3.789
3.700
4.289
2.440
-4.093
5
IPB Dramaga
-6.287
0.417
1.192
-0.717
-0.915
-6.044
6
Gunung Picung
-4.709
-17.370
2.974
-0.753
-2.992
N/A
7
Citeureup
-1.863
-0.906
4.294
5.976
6.627
1.841
8
Hambalang
5.798
-0.811
1.913
5.932
22.489
-12.132
9
Katulampa
1.705
-4.573
-2.619
-0.601
1.085
-2.728
10
Lapangan Sempur
-8.836
1.631
5.337
3.597
10.164
-0.760
11
BIG
-9.051
-1.807
0.692
3.201
2.958
-4.308
Mean
4.435
4.847
2.750
2.828
5.681
4.401
Maximum
9.051
17.370
5.337
5.976
22.489
12.132
Minimum
0.931
0.314
0.692
0.601
0.915
0.273
RMS
5.033
7.205
3.008
3.365
8.002
5.618
It is interesting that the best mean deviation which indicate the
DEM-accuracy is achieved by TerraSAR-X data, while the
worst is SRTM90. However, in all DEM there is extrem
deviation, for instance 17.370 m in IFSAR DTM or 22.489 in
SRTM. The extreme value found in the data SRTM90
Hambalang and Field Sempur, DEM RBI (Hambalang) and
DTM (Sentul and Gunung Picung). The points of difference
between the value has more than 10m with GPS measurement
results, althoug the site of test points are normally flat enough.
There are several factors that affect the accuracy of DEM, but in
this discussion will be emphasized two things, the resolution
and the shape of the earth's surface (which is represented in the
tilt / slope).
Resolution is the main factor that makes precision SRTM90 and
DEM RBI is relatively low. SRTM90 has a spatial resolution of
90m, while the RBI DEM about 100m. DEM RBI shaped
contour with the method of interpolation triangulated Irregular
Network (TIN), which is then converted to raster using tools
TIN to Raster, so that the resulting resolution is the default of
these tools.
As described in Chapter 1, the difference in resolution makes
topography at each DEM data is to be different. The higher the
spatial resolution of a DEM data, then the profile will be more
detailed represented, so the level of elevation error will be
smaller. Here is a different profile of the topography in the area
around the point Hambalang data TerraSAR-X, SRTM90, RBI
DEM and DTM.
This contribution has been peer-reviewed.
doi:10.5194/isprsarchives-XLI-B4-111-2016
115
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLI-B4, 2016
XXIII ISPRS Congress, 12–19 July 2016, Prague, Czech Republic
4. CONCLUSSION
The results shown that DEM from TerraSAR-X and SRTM30
have the best quality. Based on this research, it was inferred
that these parameters are still positively correlated with the
basic concept, namely that the lower and the higher spatial
resolution topography of an area of DEM data, the more
imprecise the resulting vertical height.
(a)
ACKNOWLEDGEMENTS
(b)
Fig 3. Difference of of DEM data from (a) SRTM90 (b) RBI
Fourth profiles were taken on the same segment by segment
length + 100m. It can be seen that the difference between
TerraSAR-X profile with SRTM90 and DEM RBI so great,
even if the SRTM90 and DEM RBI profile relative decline, in
the TerraSAR-X profile tends to rise. The y-axis of the profile
can also show the level of "error" and DEM SRTM90 RBI,
where the maximum height of the TerraSAR-X on that profile
only + 452m, whereas in SRTM90 achieve and DEM + 469m +
459m RBI reached.
The author will thanks to Agung, Danang, Irwan, and Dadan for
the generous help in the GPS measurement and its processing,
to Dr. Sri Hartini for the compilation, Dr. Ade Komara Mulyana
for some idea and data set, and to Ms Niendyawaty in Center
for Research, Promotion and Cooperation BIG for their
financial and administrative support.
REFERENCES
Athmania, D., et al., (2014). External Validation of the ASTER
GDEM2, GMTED2010 and CGIAR-CSI- SRTM v4.1 Free
Access DEMs in Tunisia and Algeria. Remote Sensing.
(www.mdpi.com/journal/remotesensing)
Forkuor, G.et al. (2012). Comparison of SRTM and ASTER
Derived Digital Elevation Models over Two Regions in Ghana.
Implications for Hydrological and Environmental Modeling. In
Studies on Environmental and Applied Geomorphology;
Piacentini, T., Ed.; InTech: Rijeka, Croatia; pp. 219–240.
Gesch D., (2005). Vertical Accuracy of SRTM Data of the
Accuracy of SRTM Data of the United States: Implications for
Topographic Change Detection, SRTM Data Validation and
Applications Workshop.
(a)
(b)
Fig 4. Topographic profil on data (a) SRTM30, (b) DTM
ASTER GDEM and SRTM30 which has the same spatial
resolution and the error rate is almost the same at that point has
a height range which does not differ with TerraSAR-X. The
maximum height of the ASTER GDEM is + 452 m, while
SRTM30 is + 453 m. DTM from TerraSAR-X has a resolution
of 10m, and a maximum height of the DTM is + 449 m. Those
values are still relevant with regard to the resolution of the
DEM used, with reference to the sample points that have
relatively low accuracy and the topography is relatively high.
Subsequent analysis is an analysis based on the height of the
area. Value of the difference with a value of> 10m generated in
the DTM (Sentul and Gunung Picung), SRTM90 (Hambalang
and Field Sempur), and DEM RBI (Hambalang). Based on the
height, just point Sempur Field which has an altitude below
300m, and a large difference can be caused by SRTM90
relatively low resolution. Based on this research, it was inferred
that these parameters are still positively correlated with the
basic concept, namely that the lower and the higher spatial
resolution topography of an area of DEM data, the more
imprecise the resulting vertical height.
Hoja D. and d’Angelo P, (2010). Analysis Of DEM
Combination Methods Using High Resolution Optical Stereo
Imagery And Interferometric SAR Data, International Archives
of the Photogrammetry, Remote Sensing and Spatial
Information Science, Volume XXXVIII, Part 1, Calgary,
Canada.
Li, P. et al., (2013). Evaluation of ASTER GDEM using GPS
benchmarks and SRTM in China. International. Journal of
Remote Sensing, 34, 1744–1771.
Li, Z., Zhu, Q., and Gold, C., (2005). Digital Terrain Modeling
Principles & Methodology. CRC Press. Florida. USA.
Mukherjee, S. et al., (2013). A. evaluation of vertical accuracy
of open source Digital Elevation Model (DEM). International.
Journal. Application. Earth Observation. Geoinformation. , 21,
205–217.
NASA, (2005). An Assessment of the SRTM Topographic
Products. NASA. Amerika Serikat.
Petrie, G. and Kennie T., (1987). An introduction to terrain
modeling: applications and terminology. Univ of Glasgow.
Raggam, H., et al (2010). Accuracy Assessment of 3D Point
Retrieval From Terrasar-X Data Sets. Proceeding of 8th
European Conference On Synthetic Aperture Radar 2010
This contribution has been peer-reviewed.
doi:10.5194/isprsarchives-XLI-B4-111-2016
116